Drs. Kinloch and Venables are co-inventors on a patent for the anti–citrullinated α-enolase peptide 1 antibody (patent no. 08701940).
Immunization with Porphyromonas gingivalis enolase induces autoimmunity to mammalian α-enolase and arthritis in DR4-IE–transgenic mice
Article first published online: 29 NOV 2011
Copyright © 2011 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 63, Issue 12, pages 3818–3823, December 2011
How to Cite
Kinloch, A. J., Alzabin, S., Brintnell, W., Wilson, E., Barra, L., Wegner, N., Bell, D. A., Cairns, E. and Venables, P. J. (2011), Immunization with Porphyromonas gingivalis enolase induces autoimmunity to mammalian α-enolase and arthritis in DR4-IE–transgenic mice. Arthritis & Rheumatism, 63: 3818–3823. doi: 10.1002/art.30639
- Issue published online: 29 NOV 2011
- Article first published online: 29 NOV 2011
- Accepted manuscript online: 27 SEP 2011 10:22AM EST
- Manuscript Accepted: 18 AUG 2011
- Manuscript Received: 6 APR 2011
To examine the hypothesis that the subset of rheumatoid arthritis (RA) characterized by antibodies to citrullinated α-enolase is mediated by Porphyromonas gingivalis enolase in the context of DR4 alleles.
Recombinant human α-enolase and P gingivalis enolase, either citrullinated or uncitrullinated, were used to immunize DR4-IE–transgenic mice and control mice (class II major histocompatibility complex–deficient [class II MHC−/−] and C57BL/6 wild-type mice). Arthritis was quantified by measurement of ankle swelling in the hind paws and histologic examination. Serum IgG reactivity with α-enolase and citrullinated α-enolase was assayed by Western blotting and enzyme-linked immunosorbent assay (ELISA). Antibodies to peptide 1 of citrullinated α-enolase (CEP-1) and its arginine-bearing control peptide, REP-1, were also assessed by ELISA.
Significant hind-ankle swelling (≥0.3 mm) occurred in DR4-IE–transgenic mice immunized with citrullinated human α-enolase (9 of 12 mice), uncitrullinated human α-enolase (9 of 12 mice), citrullinated P gingivalis enolase (6 of 6 mice), and uncitrullinated P gingivalis enolase (6 of 6 mice). Swelling peaked on day 24. None of the control groups developed arthritis. The arthritic joints showed synovial hyperplasia and erosions, but there was a paucity of leukocyte infiltration. Antibodies to human α-enolase, both citrullinated and unmodified, and to CEP-1 and REP-1 were detectable in all immunized mice except the class II MHC−/− control mice.
This is the first animal model that links an immune response to P gingivalis enolase to an important subset of RA, defined by antibodies to citrullinated α-enolase in the context of DR4. The fact that arthritis and anti–CEP-1 antibodies were induced independent of citrullination of the immunizing antigen suggests that the unmodified form of α-enolase may be important in initiating the corresponding subset of human RA.
Autoantibodies to citrullinated proteins are diagnostically specific for rheumatoid arthritis (RA) and may precede the onset of clinically apparent disease by many years (for review, see ref.1). Citrullinated proteins arise from the posttranslational modification of arginine, catalyzed by peptidylarginine deiminases (PADs). The presence of antibodies to citrullinated proteins, as detected by the anti–cyclic citrullinated peptide (anti-CCP) test, is now included as a criterion in the revised classification criteria for RA (2). Anti-CCP is a generic test that uses artificial peptides for detection of antibodies to several citrullinated proteins. Among these are antibodies to an established autoantigen, citrullinated α-enolase (3), whose immunodominant epitope has been mapped to peptide 1 of citrullinated α-enolase (CEP-1) (4). This peptide contains 2 citrulline residues, both of which are citrullinated by PAD-2 in vitro (3). Anti–CEP-1 antibodies are specific for RA and constitute 40–60% of anti-CCP–positive sera (5). The α-enolase protein, some of which is citrullinated as detected by an anti–CEP-1 antibody, is abundantly expressed in the inflamed joint (6), and anti–CEP-1 antibodies are preferentially concentrated in the rheumatoid synovial fluid as compared to the serum, suggesting that this is a local immune response (7).
A major susceptibility gene for RA encodes an amino acid sequence common to HLA–DR alleles, including DR4, collectively known as the shared epitope, which shows an epidemiologic interaction with the environmental risk factor cigarette smoking in producing anti-CCP–positive rheumatic disease (8). We have recently shown that the gene–environmental association is further segregated to the subset of patients with both anti-CCP and anti–CEP-1 antibodies, suggesting that citrullinated α-enolase is a specific citrullinated autoantigen that links smoking to DR4 alleles in the development of RA (5).
Smoking is not the only environmental risk factor for the development of RA. Periodontitis, an inflammatory disease of the tissues surrounding and supporting the teeth, is more common in patients with RA, and RA is more common in patients with periodontitis (for review, see ref.9). Periodontitis, even in the majority of subjects without RA, is also linked to the shared epitope and smoking (for review, see ref.10). Porphyromonas gingivalis, a major pathogen in periodontitis, has attracted interest as a possible etiologic factor in RA, because it is the only bacterium known to express its own PAD (11). Similar to other prokaryotes, P gingivalis also expresses a bacterial enolase, which has 51.4% amino acid overlap with the human ortholog. Of note, the homology rises to 92% over the CEP-1 sequence (4). Moreover, we have recently demonstrated that P gingivalis PAD can citrullinate human α-enolase peptides (12). Therefore, it is possible that anti–citrullinated enolase antibodies in a subset of patients with RA (5) could be primed, in the context of DR4 alleles and smoking, by an immune response to P gingivalis citrullinated enolase.
DR4-IE–transgenic mice, developed by Ito et al using mice on a C57BL/6 background (13), provide an ideal model to study human DR4-restricted autoimmune responses. Hill and coworkers (14) demonstrated that immunization of these DR4-IE–transgenic mice with in vitro–citrullinated human fibrinogen induced a mild but progressive arthritis after 50 days. No arthritis occurred in the wild-type (WT) controls, nor was there any disease development in transgenic mice immunized with fibrinogen that had not been citrullinated in vitro (14). In the present study, we used DR4-IE–transgenic mice to investigate the arthritogenic effects of human α-enolase and P gingivalis enolase. Thus, this study provides a new model of tolerance bypass by a bacterial antigen in the context of major histocompatibility complex (MHC) alleles commonly expressed in patients with RA.
MATERIALS AND METHODS
Class II MHC–deficient (class II MHC−/−) and DR4-IE–transgenic mice (bred on the same class II MHC II−/− C57BL/6 background) were purchased from Taconic and housed as previously described (14). WT mice on the C57BL/6J background were purchased from The Jackson Laboratory. All mice were male and 8–10 weeks of age at the time of immunization. Mice were housed in the barrier facility at The University of Western Ontario, in accordance with provincial husbandry standards.
Recombinant human α-enolase was cloned, expressed, and purified according to the protocol described by our group (6). Purified antigen was subsequently dialyzed against ultrapure water and then incubated in PAD buffer (0.1M Tris HCl, pH 7.4, 10 mM CaCl2, 5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 1 μg/ml leupeptin, 10 μg/ml pepstatin), with or without rabbit skeletal PAD-2 (Sigma), to generate citrullinated and uncitrullinated antigens, respectively, as previously described (6). Antigens were concentrated using 30-kd MWCO columns (Vivascience) prior to emulsification in Freund's complete adjuvant (CFA). P gingivalis enolase was cloned and expressed using the same system (4).
Mice were immunized subcutaneously on both sides of the inner thigh/flank with a total of 100 μg of antigen that had been emulsified in CFA. Mice received a booster injection with antigen emulsified in Freund's incomplete adjuvant, as was previously described for fibrinogen, on day 21 (14). Three nonimmunized age-matched mice per group were also assessed for change in ankle thickness. The width of the ankles of any of these nonimmunized control mice did not change by any more than 0.2 mm over the course of 30 days.
Assessment of arthritis.
Both hind ankles were measured every other day for changes in thickness. The ankle thickness was measured using Vernier calipers (VWR Scientific Products), as performed previously by Hill et al (14).
Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting.
Citrullination of antigens was confirmed by 3 different methods. Native and citrullinated antigens were resolved by SDS-PAGE (precast 4–12%, 15-well, 1.5 mm–thick Bis-Tris Minigels; Invitrogen) and then either stained with instant blue (Triple Red Laboratory Technology) to visualize an accelerated mobility (in the case of citrullinated enolase) or electrotransferred to nitrocellulose for immunoblotting, using an anti–modified citrulline kit (Upstate Biotechnology) or a rabbit polyclonal antibody raised against CEP-1 (preferential for the citrullinated form of the CEP-1 epitope), as described previously (4, 6). Rabbit polyclonal anti–CEP-1 binding was detected using a goat anti-rabbit horseradish peroxidase secondary antibody (Dako) at 1:5,000 in blocking buffer (as described below).
Mouse sera were tested for the presence of IgG antibodies to uncitrullinated and citrullinated α-enolase, by immunoblotting 200 ng/lane of resolved antigen (prepared as described above). Membranes were blocked with blocking buffer (5% skim milk powder in phosphate buffered saline [PBS]–0.1% Tween) for 3 hours. Subsequently, they were blotted with mouse serum, diluted 1:50 in blocking buffer overnight at 4°C, and washed in PBS–0.1% Tween. The α-enolase was detected with a polyclonal rabbit anti-enolase antibody (Santa Cruz Biotechnology) at a dilution of 1:1,000. Antigen-specific antibodies in murine sera were detected following 1 hour of incubation with biotin-conjugated anti-mouse IgG (diluted 1:1,000 with blocking buffer; Sigma-Aldrich), followed by washing in PBS–0.1% Tween, with results assessed using enhanced chemiluminescence (GE Healthcare).
Enzyme-linked immunosorbent assays (ELISAs).
Relative binding of IgG antibodies to each of the antigens (citrullinated and uncitrullinated human α-enolase and P gingivalis enolase) and peptides (CEP-1 and its arginine-bearing control peptide REP-1) was measured by ELISA, adapting a CEP-1 ELISA protocol that has been previously reported (4). Plates were coated with 20 μg/ml (50 μl/well) antigen or 10 μg/ml (50 μl/well) peptide in coating buffer, blocked with 5% bovine serum albumin in PBS for 5 hours, and incubated with mouse serum that was diluted 1:10 overnight at 4°C in radioimmunoassay buffer containing 10% fetal calf serum. Plates were washed 5 times in PBS–0.1% Tween. Incubations with anti-IgG were performed by diluting the serum to a final concentration of 1:5,000 and then adding 50 μl/well anti-IgG, followed by incubation by shaking at room temperature for 1.5 hours. All ELISAs were developed with tetramethylbenzidine (KPL), and the optical density (OD) at 450 nm was read. All OD values are given following the subtraction of background (coating buffer alone). When the subtracted values were less than zero, a value of zero was given to the respective samples.
Mice were killed on day 30, and the hind legs were removed, stored in buffered 10% formalin, and decalcified. The leg joints were then sectioned and stained with hematoxylin and eosin, as described previously (14).
Data analyses were performed using GraphPad Prism version 4.0. Statistically significant differences in ankle swelling at the respective time points were calculated using two-way analysis of variance and Bonferroni post hoc tests. Statistically significant differences in antibody titers were determined using the Mann-Whitney U test. The percentage of amino acid overlap of enolase orthologs was calculated using the alignment program on the Expasy Web site (http://.expasy.org/cgi-bin/sim.pl).
Rapid onset of arthritis induced by human α-enolase in DR4-IE–transgenic mice.
DR4-IE–transgenic mice were immunized with purified human α-enolase (details available from the corresponding author upon request), using the same protocol as has been described for citrullinated fibrinogen (14). Nine of the 12 mice immunized with citrullinated enolase developed arthritis in both hind paws, with the mean increase in ankle swelling peaking on day 24 (Figures 1A and B). Although the erythema subsided, the swelling of the ankle persisted in all arthritic mice (Figure 1B). Histologic examination showed synovial proliferation, a relatively scattered mononuclear cell infiltrate, erosions, and joint destruction affecting the ankle joint (Figure 1C) and, in some cases, affecting joints within the tarsus. Unexpectedly, the uncitrullinated antigen produced similar results. Nine of the 12 mice immunized with the uncitrullinated antigen developed arthritis, and there was no statistically significant difference in the rate or magnitude of ankle swelling between the 2 immunized DR4-IE–transgenic mouse groups (P > 0.05) (Figure 1A) (details available upon request). There was no arthritis in the control groups, comprising the C57BL/6 class II MHC−/− mouse strain and the WT C57BL/6 mouse strain. All mice were male and of the same age, to control for sex and age differences.
Development of an antibody response to human α-enolase in both arthritic mice and nonarthritic WT controls.
Following immunization, serum from each of the mice was tested on day 30 to determine the citrulline specificity of the antibody response, and to determine whether citrullination of the immunizing antigen is required to elicit a response. As detected by immunoblotting, IgG antibodies reacting with both citrullinated and uncitrullinated α-enolase were detectable in all immunized groups (Figure 2A), except in the immunodeficient class II MHC−/− mice. Similar results were also obtained in the anti–CEP-1 ELISA, which showed that antibodies to CEP-1 were induced in both WT and transgenic mice, irrespective of citrullination of the immunizing antigen (Figure 2B). However, unlike humans with RA, all mice with antibodies to CEP-1 (CKIHA-cit-EIFDS-cit-GNPTVEC) also had antibodies to REP-1 (CKIHA-arg-EIFDS-arg-GNPTVEC), the arginine-containing control peptide. As can be seen in Figure 2B, there was a greater relative binding to REP-1, indicating that the IgG response was not citrulline-dependent.
To address the possibility that epitopes other than CEP-1 were dominant during an immune response, we also developed and validated an ELISA using whole recombinant enolase (details available upon request). The results of this assay confirmed an antibody response to both forms of enolase in all immunized groups and the lack of response in the class II MHC−/− controls (Figure 2C). Although there was a trend toward greater antibody binding to the uncitrullinated antigen in both the nonimmunized and immunized groups, this difference did not reach statistical significance (P > 0.05).
Induction of arthritis and IgG antibodies cross-reactive with the CEP-1 peptide and whole human α-enolase following immunization with P gingivalis enolase.
Given the sequence similarity between human α-enolase and P gingivalis enolase, and having established that human α-enolase induces arthritis in DR4-IE–transgenic mice, we wanted to determine whether the previously described links between RA and periodontitis (15) could be attributable to an immune response to P gingivalis enolase, in the context of HLA–DR4 (Figures 3A and B). Twelve DR4-IE–transgenic mice were immunized with recombinant P gingivalis enolase (6 were immunized with the citrullinated antigen, and 6 were immunized with the uncitrullinated antigen) using the same protocol as was used for immunization with human α-enolase.
All mice developed a pattern of arthritis (Figure 3A) similar to that induced by the human ortholog. The pattern of disease development was not significantly altered by citrullination of the immunizing antigen (details available upon request). All mice developed antibodies reactive with human α-enolase protein and peptides (Figure 3B).
These results demonstrate that immunization of mice with enolase from a candidate bacterium causes arthritis and an autoimmune response similar to that seen in a corresponding subset of human RA. The results of this study provide further evidence in support of the gene–environment autoimmunity paradigm, which was previously demonstrated, by our group and others, in studies assessing the interactions between DR4, smoking, and anti–citrullinated enolase antibodies (5). In the present study, the environmental factor is P gingivalis, for which smoking may be an indirect marker, because smoking is also independently associated with P gingivalis infection (10). A surprising finding was that both uncitrullinated enolase and citrullinated enolase were immunogenic and caused arthritis. This is distinct from the findings of a previous study in which citrullinated fibrinogen was assessed in the same transgenic mouse strain (14), the results of which showed that the development of arthritis was entirely dependent on in vitro citrullination of the antigen. However, we cannot be certain that ours is a truly citrulline-independent model, because it is possible that some of the enolase was citrullinated at the site of injection or within the inflamed synovium by mouse PADs.
The enolase model also differs in that the onset of arthritis was much more rapid than that observed in the fibrinogen model. We suggest that this may be related to low levels of autoantibodies to unmodified α-enolase in 4–15% of the healthy human population (3, 10), which may be due to a preexisting and cross-reactive immune response to bacterial enolases. Such a phenomenon may also occur in mice. Preliminary evidence may be found in our results showing that the “background” binding of antibodies to REP-1 was slightly higher than that to CEP-1 in the treatment-naive (nonimmunized) mice (Figure 2B). This prior sensitization to enolases may also explain the extremely rapid onset of arthritis observed in our study.
The precise mechanisms of how such low levels of autoimmunity to α-enolase could lead to the evolution of a pathogenic autoimmune response to the citrullinated antigen in human disease are yet to be explored. However, our study clearly demonstrates that when investigating the etiology of a disease as heterogeneous as RA, it is important to focus on subsets defined by specific autoimmune reactions in the context of the appropriate susceptibility genes. For α-enolase and RA, it is the unmodified autoantigen (and bacterial orthologs) that is the key to understanding early events in disease etiology.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Venables had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Kinloch, Alzabin, Brintnell, Wilson, Wegner, Bell, Cairns, Venables.
Acquisition of data. Kinloch, Alzabin, Brintnell, Wilson, Barra, Bell, Cairns.
Analysis and interpretation of data. Kinloch, Alzabin, Brintnell, Wilson, Bell, Cairns, Venables.